Predicting Likely Receivers Throughout an NFL Play

Zachary Pipping, Lou Zhou | Karim Kassam

Motivation

  • Determining who the likely recipient can be key in making the optimal decision for the defense as well as evaluating the decision-making of a quarterback

    • Rewarding quarterbacks for finding uncommon but high-value passes
    • For a defense, determining likely throw target for appropriate positioning
  • Look to build a ranking model which determines the most likely receiver at a frame given throw attempt

Data Overview

  • 2025 NFL Big Data Bowl – Weeks 1–9
  • Game Data – Home and Away Team, Final Score, Game Time
  • Play Data – Play Description, Game Context, Play Result, Changes in Win Probability
  • Player Play Data – Statistics for each player for a play
    • Route ran by player, Whether the player made a tackle or interception
  • Tracking Data - Locations of players and the football at each frame of a play

Spacing Tells an Incomplete Story

Methodology

  • Building a ranking algorithm(e.g. XGBoost) to rank the likeliest recipient at a frame
    • Similar approaches in soccer1
    • Extracting features from tracking and play-by-play data
  • Potential pre-snap work, predicting the most likely target given the receiver alignments

Next Steps

  • Extracting features used to build the ranking model
    • Distance, Relative Speed, Relative Orientation From Nearest Defender
    • Quarterback Position, If Under Pressure
    • Number of Defenders between Quarterback and Receiver, Passing Angle
    • Game Score, Time Remaining
  • Quantifying potential of separation